Received: 18 December 2017 Revised: 29 April 2018 Accepted: 18 May 2018

DOI: 10.1002/csr.1639

RESEARCH ARTICLE

Corporate social responsibility and strategic company

behaviour: CVS Health's discontinuation of tobacco products
Lirong Liu1 | Shiyou Li2 | Michael Opara2

1
Economics and Finance, Texas A&M
University Commerce, Commerce, Texas, USA Abstract
2
Accounting, Texas A&M University The elimination of tobacco products by CVS Health represents a special case of
Commerce, Commerce, Texas, USA
corporate social responsibility (CSR) with substantial costs—about $2 billion revenue
Correspondence
Michael Opara, Texas A&M University
loss in tobacco annually. This paper examines how CVS Health's stock market perfor-
Commerce, Accounting, Commerce, Texas, mance was affected by two events: the announcement (5 February 2014) and the
USA
Email: [email protected]
implementation (1 October 2014) of the elimination of tobacco products. The study
event approach and regression models are used to examine both short‐ and long‐term
effects. The results show positive and significant abnormal returns during both
events, with effects being much stronger during the implementation event period.
In the long term, the net effects become neutral, most likely because the positive
effects of CSR and the negative effects of the revenue loss offset each other.
The results have important implications for both the retail pharmacy industry and
companies in other industries. Socially responsible actions, such as the elimination
of harmful products could impose substantial revenue losses or extra expenses, but
more importantly, they can have a positive and significant impact on corporate
financial performance. Thus, such actions benefit not only the stakeholders, but also
the shareholders, and are consistent with both stakeholder theory and shareholder
theory.

KEY W ORDS

abnormal return, corporate financial performance, corporate social responsibility, corporate stock
performance, corporate strategic behaviour

1 | I N T RO D U CT I O N societal health. This study examines the effects of this CSR action
on CVS’ stock performance.
On 5 February 2014, CVS Health (CVS) announced that it would dis- The study of CSR has been on the rise in recent literature. Frynas
continue the sale of cigarettes and tobacco‐related products from its and Yamahaki (2016) provide a comprehensive review of the theories
7850 stores in the USA. Acknowledging that selling tobacco in a retail on internal and external drivers of CSR from a theoretical perspective.
pharmacy environment conflicted with the delivery of healthcare ser- The most commonly used theory for external drivers is stakeholder
vices and with the company's stated purpose of helping people to bet- theory, while agency theory is usually adopted for internal drivers
ter health, CVS closed its tobacco division and stopped selling (Frynas & Yamahaki, 2016). Application of these theories holds a
tobacco‐related products in its stores on 1 October 2014. In making mostly positive view on CSR, suggesting that CSR can benefit corpo-
this strategic choice, CVS lost annual tobacco revenue of $2 billion rations. However, establishing the link between CSR and corporate
(Merlo, 2014). The action taken by CVS represents a unique corporate stock/financial performance though empirical means remains an unre-
social responsibility (CSR) action because it involves substantial costs, solved issue (Ducassy, 2013; Fifka & Berg, 2014; Huang & Watson,
uncertain outcomes for the company, and significant benefits to 2015; Marti, Rovira‐Val, & Drescher, 2015; Tenbtermsel, 2007). One

Corp Soc Resp Env Ma. 2018;1–13. wileyonlinelibrary.com/journal/csr © 2018 John Wiley & Sons, Ltd and ERP Environment 1
2 LIU ET AL.

stream of prior literature documents a positive relationship between cumulative abnormal returns around the announcement date
CSR and stock performance, though the size of this relationship has (5 February 2014) and the implementation date (1 October 2014).
become smaller in recent years (Huang & Watson, 2015; Margolis, The results show significant and positive impacts of the CSR action
Elfenbein, & Walsh, 2009); a second group indicates that stock perfor- on CVS stock during these two events in the short term, with impacts
mance is negatively associated with CSR, and the relationship dimin- being stronger and more significant during the implementation event
ishes (Huang & Watson, 2015). The final view contends that there is window. We also find limited evidence of negative and significant
no statistically significant relationship between CSR and stock returns abnormal returns on the day of the announcement, implying that
(Huang & Watson, 2015; Lys, Naughton, & Wang, 2015). the market was more concerned about the potential revenue loss. In
While prior literature provides mixed results, little research has our long‐term analysis, we examine monthly returns after the
been done on the effects of a CSR action that involves a substantial implementation. Both the event study and regression analysis show
potential revenue loss but significant benefit to the society. On one that the effects of the CSR on CVS’ stock market performance are
hand, the elimination of tobacco products implies an annual revenue neutral in the long term.
loss of $2 billion for CVS. This can impose detrimental effects on the The strategic action taken by CVS involves the significant revenue
company's stock price. On the other hand, such a CSR action can loss of $2 billion. This can impose detrimental effects on the
enhance a firm's reputation, attract customers and investors, and pos- company's stock price. Thus, on the surface, it seems that CVS’ action
itively affect the stock performance of CVS (Jones, 1995). Thus, the departs from shareholder theory and is more consistent with
stock market reactions to CVS’ CSR action are ex‐ante and unclear. stakeholder theory. Shareholder theory states that managers should
The purpose of this paper is to investigate the net effects of this behave in the interest of the shareholders. This theory seems to
unique CSR action on CVS’ stock performance. We consider both conflict with CSR, which is ostensibly an effort to address the needs
the short‐ and the long‐term effects. The short‐term analysis is more of non‐shareholders (Huang & Watson, 2015). The stakeholder theory
common in event studies of financial performance. This allows us to demonstrates that the needs of all stakeholders must be considered in
examine the immediate reactions to the CSR action in the financial corporate decisions (Carroll, 1991). Our empirical results suggest that
market. Given that post‐event data were available for two years at CVS’ CSR action benefits both non‐shareholding stakeholders and
the time this study was conducted, we were also able to examine shareholders. Therefore, the benefits and interests of shareholders
the long‐term effects of the CSR action on CVS’ stock performance. and stakeholders are not necessarily in conflict (Barnett & Salomon,
Since the huge revenue loss is realized and factored into the daily 2012; Garriga & Mele, 2004).
operation of the company, the effects in the long term can be different To the best of our knowledge, this is the first empirical study that
from those in the short term. Therefore, it is important to examine examines the stock market's reactions to the discontinuation of
how the financial market adjusts in the long term. tobacco products by CVS. The results of our study present a number
The importance of our study also lies in the tremendous impact of of important contributions. First, from a theoretical standpoint, this
the tobacco withdrawal by CVS on society. Smoking has long been study adds to the growing CSR literature by documenting the impact
known to be connected with serious diseases. According to the Cen- of a unique CSR action. By reviewing the stock market reaction to
ters for Disease Control (CDC), smoking causes cancer, heart disease, CVS’ corporate decision about tobacco discontinuation, this study
stroke, lung disease, diabetes, and chronic obstructive pulmonary dis- intends to identify which direction the effects of the CVS decision
ease (Centers for Disease Control and Prevention, 2017). More than may lean, and thus extend our knowledge about the outcomes of cor-
16 million Americans are living with a disease caused by smoking. porate social decisions. Our study provides empirical evidence focus-
Smoking is responsible for more than 480 000 deaths per year in ing specifically on the stock market reaction to a major CSR
the USA, including more than 41 000 deaths resulting from second‐ undertaking via harmful product withdrawal. It points to the CSR
hand smoke exposure. The CDC documents that if smoking continues implications in a setting where the cost of a CSR action is significant
at the current rate, 5.6 million of today's Americans younger than and known ab initio (a $2 billion revenue loss), and where a clear set
18 years of age are expected to die prematurely from a smoking‐ of benefits are unknown or unclear.
related illness. This, in our view, provides the context and justification Second, from a practical standpoint, this study has significant
for CVS’ withdrawal of tobacco products, and CVS’ continued focus implications for practitioners, especially corporate managers who
on helping its customers stop smoking on their paths to better health. make CSR decisions. Our study can help management make informed
The tobacco ban, together with the smoking cessation support, can decisions when contemplating a range of possible CSR alternatives.
have a fundamental impact on how we perceive and deal with The managerial choices around CSR were taken by CVS in the midst
smoking in society. It also implies potentially substantial economic of known significant costs but uncertain benefits for the corporation
savings in healthcare expenditure. Upon its announcement, CVS and shareholders. Examining this event enables corporate managers
branded its tobacco withdrawal as being socially responsible; it was to make informed choices when they consider CSR types that could
generally well received and understood by the public, as witnessed impose substantial costs (via revenue loss) on their shareholders but
by extensive positive reviews. could be beneficial to the entire community despite these costs,
We use both the event study approach and the panel data including their shareholders. Our empirical results indicate that a
regression approach to examine the CVS stock performance in the CSR action, even one with a substantial cost—a revenue loss of $2
short term and the long term. In our short‐term analysis, we consider billion in tobacco—can still have a positive impact on stock perfor-
daily returns and examine the significance of abnormal returns and mance and thus benefit both the shareholders and stakeholders. In
LIU ET AL. 3

particular, such results can help companies to make better strategic reach of CSR actions have continued to expand. For instance, CSR
decisions about future CSR plans and their competitiveness in the has extended to cover areas such as responsible marketing, product
industry. For pharmacies that are still associated with tobacco sales, responsibility and sustainable consumption. Responsible marketing
the findings of the positive effects can guide them through future incorporates the idea that businesses have a responsibility to their
decisions about banning tobacco. Even for pharmacies without customers to ensure responsible behaviours that guarantees the best
tobacco products, such results still can help them to better frame possible consumer satisfaction, innovation in products, and tangible/
CSR actions that may require significant additional costs to the measurable long‐term value proposition. Some attributes of socially
company. responsible marketing include product safety, transparency, ethical
Third, research has long attempted to uncover the relationship pricing, and respect for customer privacy. According to McWilliams,
between CSR action and financial performance (Fifka & Berg, 2014; Siegel, and Wright (2006), some analysts distinguish between per-
Flammer, 2015; Frynas & Yamahaki, 2016; Margolis & Walsh, 2003; suasive CSR advertising and informative CSR advertising. Persuasive
Orlitzky, Schmidt, & Rynes, 2003); this study adds to that effort, and CSR advertising seeks to positively influence consumer preferences
within that context engages a diversity of methods and horizons and tastes for products with CSR attributes. Ultimately, customers
towards a better exploration of the relationship. In that perspective, could perceive either persuasive and/or informative CSR advertising
our study extends the literature by not only providing further evidence as a signal of product quality and corporate image (Jahdi & Acikdilli,
of the positive effects of CSR on corporate financial performance, but 2009; Oberseder, Schlegelmilch, & Murphy, 2013). Servaes and
also by examining a significant and unique CSR event. On the policy Tamayo (2013) document that consumers may value CSR more
front, the withdrawal of a harmful product by CVS adds a new dimen- when the product quality is high. Contrasting experiments suggests
sion that policymakers can consider, should CSR legislation become that consumers may also value high quality or luxury goods less
necessary. It is our view that a recognized harmful product merits when they are aware of CSR, as it may reduce their assessment
legislative or policy action, just like toxic waste and other identified and perception of the brand's exclusivity (Torelli, Basu Monga, &
harmful goods. Kaikati, 2012).
The remainder of the paper is organized as follows: theoretical Closely related to responsible marketing and advertising is prod-
framework and hypotheses development are covered in Section 2. uct responsibility, which focuses on how a business can strategically
Section 3 discusses data and methodology. Results are presented in use the multidimensionality of CSR to enhance product quality and
Section 4, and discussions and conclusions are included in Section 5. thereby distinguish itself from its competitors, thus boosting consumer
attraction and retention (Banerjee & Wathieu, 2017). Sustainable con-
sumption and production proceed hand‐in‐hand and are complemen-
tary to sustainable development, and thus support long‐term
2 | T HE O R E T I C A L F R A M E W O R K A N D
economic growth consistent with environmental and social needs
H Y P O T H E S E S D E V EL O P M EN T
(OECD, 2008). As pointed out in Beare, Buslovich, and Searcy
(2014), by using the instrumentality of regulations and taxes, govern-
2.1 | Context ments can mitigate the environmental impacts of unsustainable pro-
Theorization of corporate social responsibility (CSR) has been on the duction practices and promote sustainable consumption. In addition,
rise in business literature during recent decades (Frynas & Yamahaki, there is also an increased interest in CSR in the academic community
2016; Huang & Watson, 2015). For instance, a shareholder perspec- in other areas (De Bakker, Groenewegen, & Den Hond, 2005).
tive of corporations viewed CSR as an external threat to their business Analysts have called for CSR to be better linked to overall corporate
practices and profits in the 1970s (Friedman, 1970; Hoffman & strategies (Orlitzky et al., 2011; Porter & Kramer, 2006), sensitive to
Ventresca, 2002). However, through a combination of influential cultural differences (Miras‐Rodriguez, Carrasco‐Gallego, & Escobar‐
global actors, the redefinition of the role of government, the rise of Perez, 2015), and linked to the preferences of stakeholders (Michelon,
social movements, popular legislative action, and growing public atten- Boesso, & Kumar, 2013).
tion, CSR and its variants have emerged as an organizational routine Meanwhile, the proliferation of CSR action has been paralleled by
activity or “taken‐for‐granted” (Zucker, 1983, 1987) strategic consid- growth in CSR ratings and rankings (Porter & Kramer, 2006). The
eration for major businesses. As public CSR awareness grew, the Economist (2008) notes that 39.5% of executives currently give CSR
norms of corporate social practice changed. Accordingly, governments high priority, while three years ago it was only 22.8%. In addition,
have enacted several health, safety, and environmental regulations. 16.7% of executives currently give CSR a very high priority, while
Furthermore, the UN has established global treaties on health, safety three years ago it was only 11.3%. All of this suggests that corporate
and environmental issues such as toxic chemical controls, hazardous social responsibility has moved from the margins into the centre of
waste shipments, pesticide use, and global climate change protocols. corporate activity (Landrum & Ohsowski, 2018).
In our view, tobacco withdrawal by CVS represents a socially respon-
sible action in promoting public health.
2.2 | CSR and short‐term financial performance
CSR action has become the norm in several industries, as share-
holders demand progress in addressing identified health, safety and One of the most significant theories of CSR is the stakeholder theory,
environmental issues as part of a corporate social responsibility strat- which states that the needs of all stakeholders must be considered in
egy (Orlitzky, Siegel, & Waldman, 2011). Correspondingly, the role and corporate decisions (Carroll, 1991). The theoretical foundation/basis
4 LIU ET AL.

for the normative stakeholder model is that businesses are custodians Hypothesis 2. A firm's socially responsible activities
of our collective social values, or social moral agents, entrusted with positively affect the short‐term share price performance
direct obligations to the society (Donaldson & Preston, 1995; Gibson, of the organization.
2000). Proponents of stakeholder theory argue that CSR investments
have a positive impact on corporate performance (Freeman, Harrison, 2.3 | CSR and long‐term financial performance
& Wicks, 2007; Freeman, Harrison, Wicks, Parmar, & de Colle, 2010).
While the relationship between CSR and short‐term financial perfor-
For instance, CSR allows a firm to better market itself, its products and
mance (abnormal returns) has received substantial research attention
services, to enjoy enhanced brand image and good reputation
(Frynas & Yamahaki, 2016; Gomez‐Bezares, Przychodzen, &
(Fombrun, 1996). Furthermore, the stakeholder perspective suggests
Przychodzen, 2016; Huang & Watson, 2015; Vogel, 2005), the same
that CSR investments stabilize inherent conflicts between various
cannot be said of long‐term financial performance. We argue that
interests within the organization (Freeman et al., 2007; Freeman
the theories discussed previously still apply to the long‐term analysis.
et al., 2010). In branding its product withdrawal as a special case of
If a firm can benefit from CSR in the short term, then it is likely that
CSR, CVS CEO Larry Merlo attempted to socially reposition CVS,
such advantages can continue on in the long term. Prior research sug-
redefine the social norms of the industry and attract social approval
gests that CSR is positively related to firm market value – a proxy for
(legitimacy) to CVS. Thus, it is possible that the tobacco ban can
long‐term performance. According to Kim (2010), who studied long‐
improve CVS’ financial performance through appealing to the public,
term financial performance using archival data on composite CSR
one of the stakeholders.
scores and individual CSR dimensions, CSR is negatively related to
Empirical studies of CSR and financial performance can also pro-
both a firm's systematic and unsystematic risks; suggesting that once
vide insights into CVS’ action. Ghoul, Guedhami, Kwok, and Mishra
recognized as an ethical company, a firm can reduce its risks. Impor-
(2011) find that the cost of equity increases for firms in the USA that
tantly, Brammer and Millington (2008) explored the relationship
participate in the two “sin” industries, tobacco and nuclear power.
between corporate social performance and financial performance
Given that the action taken by CVS is to pull the product out from
using charitable giving as a specific context. They found that poor
one of the sin industries, CVS may gain from positively appealing to
social performing firms did best in the short term, and good social
the public. Hong and Kacperczyk (2009) study the “sin stocks”
performing firms did best in the long term. Furthermore, Peters and
(traded stocks of tobacco, alcohol and gaming companies), and find
Mullen (2009) tested the cumulative effect of CSR on financial perfor-
that they are not as widely held by norm‐constrained investors such
mance using a time‐series approach and found that the effects of CSR
as pension plans compared to mutual/hedge funds. This implies that
on firm financial performance are positive and strengthen over time.
stocks of entities like CVS would become more attractive to institu-
This result supports the view that long‐term CSR is positive for both
tional investors such as pension funds and similarly norm‐constrained
shareholders and stakeholders.
investors.
Meanwhile, several studies that explore the relationship between
Similar positive effects of CSR are also documented by other
social responsibility and accounting‐based performance measures
studies. Using KLD data of 118 companies, Michelon et al. (2013) doc-
have produced mixed results. Cochran and Wood (1984) located a
ument that companies with CSR linked to their stakeholder prefer-
positive correlation between social responsibility and accounting per-
ences experience a strengthened company performance in both
formance after controlling for the age of assets. Aupperle, Carroll,
market‐based and accounting‐based performance measures (see also
and Hatfield (1985) detected no significant relationship between
Marti et al., 2015). Becchetti, Ciciretti, Hasan, and Kobeissi (2012)
CSR and a firm's risk‐adjusted return on assets. In contrast, Waddock
examine the corporate entry and exit from the Domini 400 Social
and Graves (1997) found positive and significant relationships
Index and find negative effects on abnormal returns after the exit.
between an index of CSP and performance measures, such as Return
While investigating whether CSR performance affects information
on Assets (ROA) in the following year. Given the mixed results from
asymmetry, Cho, Lee, and Pfeiffer (2013) find that CSR performance
the long‐term studies, we propose the following alternative hypothesis:
can be important for investors as it reduces information asymmetry.
These findings support the argument that firms with socially responsi- Hypothesis 3. A firm's socially responsible activities
ble practices have higher value and lower risk. Furthermore, Clarkson, positively affect the long‐term share price of the
Fang, and Richardson (2013) document that voluntary environmental organization.
disclosures serve as a signal to investors, which enhances stock price
and firm value. Flammer (2015) uses CSR‐related shareholder pro-
3 | M E TH OD O LO GY A N D D A T A
posals to stimulate random assignment of CSR activities and finds that
CSR leads to positive stock returns and enhanced earnings
performance. 3.1 | Short‐term analysis
With the foregoing theoretical and empirical justifications, we
Following previous literature, we implement event studies of CVS’
propose the following hypotheses: abnormal returns, a widely accepted measure of firm stock perfor-
mance. We obtain the abnormal returns using the Fama–French
Hypothesis 1. The announcement of a firm's future
five‐factor model with the following specification:
socially responsible activities positively affects the short‐
term share price performance of the organization. Rt ¼ α þ β1 Ret þ β2 SMBt þ β3 HMLt þ β4 RMW þ β5 CMA þ εt ; (1)
LIU ET AL. 5

where Rt is the excess return (with respect to the risk‐free rate) of CVS 3.2 | Long‐term analysis
stock on day t, Ret is the value‐weighted return of all CRSP stocks,
In our long‐term analysis, we examine the monthly return of CVS
SMBt is the small‐minus‐big market capitalization factor return, HMLt
stock from October 2014 to December 2016. There are two major
is the high‐minus‐low market equity factor return, RMW is the differ-
methods in the long‐term event study, the buy‐and‐hold abnormal
ence between the average return on the two robust operating profit-
return approach (BHAR) and Jensen's alpha approach (Kothari &
ability portfolios and the average return on the two weak operating
Warner, 2007).
profitability portfolios, CMA is the average return on the two conser-
The BHAR method is a characteristic‐based matching approach, in
vative investment portfolios minus the average return on the two
which the treated unit is matched to a control unit based on company
aggressive investment portfolios, and εit is a white noise random error.
characteristics. Then the differences between the cumulative return of
We first examine the effects of both the announcement and the
the treated unit and that of the control unit over the long term are cal-
implementation of the tobacco withdrawal in the short term. To deter-
culated and the average of such differences across firms is tested for
mine the appropriate event window, we take into consideration the
significance. However, since our study is a single‐firm single‐event
possibility of information leakage and extended market reaction
case, we cannot average this across firms and thus, cannot possibly
(Barnett & King, 2008). Information leakage can lead to market reac-
obtain a standard deviation for our significance test. Furthermore, in
tion prior to the actual event day. In our case here, CVS’ intention to
long‐term studies, the abnormal return should be evaluated on the
withdraw tobacco in October was known prior to the event day since
basis of post‐event estimate, and thus the pre‐event estimation win-
CVS announced its plan the previous February. In addition, given that
dow used in short‐term analysis doesn't apply (Kothari & Warner,
the CSR action taken by CVS involves direct impact on its revenue, it
2007). We turn, therefore, to regression analysis. Instead of comparing
may take days for the market to resolve uncertainty and fully adjust to
the long‐term cumulative return of the treated and the control units,
the event. Following previous literature, we choose an 11‐day [−5, 5]
we test the average long‐term returns of CVS and its competitors to
window for both events (Howe & Schlarbaum, 1986; Kryzanowski &
examine if CVS experienced statistically significant and positive or
Nemiroff, 2001). For the estimation window, we choose a three‐
negative returns in the long term. Our regression model takes the
month window and a four‐month window, using 30 days prior to the
following format:
event as the end date of the estimation window. The five‐factor model
is estimated using the ordinary least squares (OLS) method within the
Returni;l ¼ α þ δIWalgreen þ δIRA þ βF i;l þ ϵi;l (6)
estimation window. The abnormal returns are then calculated as the
difference between the realized and predicted returns on day t in
Here Returni, l is the return of company i in month l. We create
the event period. That is,
two dummy variables to represent the two competitors, Walgreen
 
and Rite Aid, so that the coefficients of these dummy variables cap-
ARt ¼ Rt − b β1 Ret þ b
αþb β2 SMBt þ b
β3 HMLt þ b
β4 RMW þ b
β5 CMA (2)
ture the statistical difference in the monthly returns between CVS
and its competitors in the long term. Specifically, IWalgreen (IRA) is a
The test on the abnormal returns that is feasible is the traditional
dummy variable that takes a value of one if an observation belongs
t‐test. The test statistic is given by
to Walgreen (Rite Aid) and zero otherwise. Thus, the omitted category
ARt represents CVS. If the coefficient is statistically significant and positive
t¼ (3)
(negative), then the competitor experienced significant higher (lower)
;
SAR
returns than CVS after the event in the long term. Control variables
where SAR is the standard error of the abnormal return calculated represented by F i, t include the Fama–French five factors and book‐
based on the estimation window. to‐market ratio. Lastly, α and ϵi, l denote the constant and the random
We further consider cumulative abnormal returns. Our prelimi- errors, respectively. The panel data set is estimated using the pooled
nary tests show that the abnormal returns tend to be consistently pos- OLS method for the sample period after the event, October 2014 to
itive starting from day 0 of the event. Therefore, we consider several December 2016.
alternative event windows for our analysis of the cumulative abnormal The second method in long‐term event study, Jensen's alpha
returns. They are [−5, 5], [0, 5], and [0, 4]. The test statistic is given by: approach, can also be modified to accommodate our single‐firm sin-
gle‐event case. We first estimate a modified Fama–French four‐factor
regression model according to Mitchell and Stafford (2000) and Serra
CAR
t¼ ; (4) (2002) with the following specification:
SCAR

where CAR is the cumulative abnormal returns, obtained by summing Rt ¼ α þ β1 Ret þ β2 SMBt þ β3 HMLt þ β5 UMD þ εt ; (7)
up the abnormal returns over the whole event window. SCAR is the
standard error of the cumulative abnormal return calculated as: where Rt, Ret , SMBt, and HMLt are the same as previously defined, and

qffiffiffiffiffiffiffiffiffiffiffiffi UMD is the difference between the highest and lowest monthly
SCAR ¼ S2AR *T ; (5) returns in the previous 12 months. Jensen's alpha is defined as the
constant, α. It captures the monthly abnormal returns of CVS stock
where T is the number of event days. over the sample period after the event date. Thus, the significance test
6 LIU ET AL.

on alpha in the regression effectively tests the statistical significance the value‐weighted return of all CRSP stocks in the US, SMB, HML,
of the monthly abnormal returns in the long term. RMW and CMA. Furthermore, using the Mergent database, we gath-
To further verify the robustness of our results, we also estimate a ered quarterly data on the book‐to‐market ratio.
panel data fixed‐effect model on the monthly returns of CVS and its Summary statistics are shown in Table 1. The last two columns
competitors over the whole sample period from January 2013 to show the overall means and standard deviations across all companies.
December 2016. The specification of the panel data model is similar We also disaggregate the returns by companies and show the overall
to Equation (5) except that we replace the two company dummy monthly returns as well as returns before and after the implementa-
variables with a new dummy variable that takes a value of one if the tion. The overall daily return of CVS stock in the whole sample period
observation belongs to CVS after the event. is 0.052 with a standard deviation of 1.145. As shown in the second
We choose our control group based on the criteria that the com- column, the average monthly return of CVS stock is, in general, nega-
pany is a nationwide retailer with a comprehensive pharmacy as one tive (−0.017), and the average monthly return actually decreased from
of its main products and services, has similar business characteristics −0.012 to −0.021 after CVS started the tobacco ban. While the aver-
and serves the same type of customers. We also examined a list of age monthly returns of Walgreen and Rite Aid are positive, they also
CVS competitors from National Association of Securities Dealers declined after October 2014.
Automated Quotations (NASDAQ). Among the companies on the list,
most of them are healthcare companies. Since the main impact of
tobacco withdrawal is not on CVS’ pharmacy sales but rather on |
4 RESULTS
household product sales, these healthcare companies are not
appropriate for our analysis. Ultimately our control group consists of
Walgreen and Rite Aid, which engage in household products in addition
4.1 | Results of short‐term analysis
to a comprehensive pharmacy. The results of the short‐term event study are shown in Tables 2–4.
The fixed‐effects panel data model allows us to examine the We first examine the hypotheses in regard to the effects of the
effects while controlling for any industry trend in the stock market announcement on 5 February 2014. We hypothesize that the
and company‐specific characteristics. Specifically, the fixed effects announcement has a positive effect on CVS’ stock performance.
model controls for any time‐invariant company characteristics that The results of the test of the announcement are shown in
are unobservable or omitted in the model, while the industry trend Table 2. We consider an 11‐day window from five days before the
can be captured by including the competitors in the model. This allows event to five days after the event. The abnormal returns, t‐test statistics
for unbiased estimation of the model and thus an unbiased effects and the associated p‐values obtained using a three‐month estimation
estimation. In addition, this method also enables us to compare CVS window are shown in columns 2–4 while such statistics obtained using
and its competitors based on the stock market. a four‐month estimation window are shown in columns 5–7. Positive
abnormal returns are found for days −4, −3, 3 and 4, as shown in
columns 2 and 4. Based on the p‐values reported, these positive
3.3 | Data abnormal returns are significant on day 4 based on both estimation
Our data come mainly from the US Securities and Exchange Commis- windows, and significant on day 3 based on a four‐month estimation
sion (SEC) and NASDAQ, from which we obtain daily stock market window. Such findings are strong evidence in support of Hypothesis 1.
prices for CVS, Walgreen and Rite Aid for four years from 2013 to The positive abnormal returns suggest that the CSR action of CVS,
2016. Thus, we have a total of 3024 observations, or 742 for each although a risky move, has positive impacts on its stock market perfor-
of the three companies, CVS, Walgreen and Rite Aid. To implement mance. It also indicates that the positive signal of CVS being socially
the Fama–French five‐factor model to obtain the abnormal return, responsible outweigh the concerns about the potential revenue loss.
we gathered data from the homepage of Kenneth French, including According to a 2015 report by the Arthur Page Society, CVS was

TABLE 1 Summary statistics

CVS Walgreen Rite Aid Overall
Variable Mean Std. Dev. Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.

Overall Daily Returns 0.052 1.145

Overall Monthly Return −0.017 0.046 0.014 0.697 0.035 2.487 0.011 0.100
Before Implementation −0.012 0.048 0.021 0.085 0.063 0.158 0.024 0.110
After Implementation −0.021 0.045 0.009 0.057 0.014 0.140 0.001 0.091
Book‐to‐market ratio 20.435 14.587
Market Return (Ret ) 0.052 0.830
Small Minus Big (SMB) 0.001 0.483
High Minus Low (HML) 0.012 0.453
Robust Minus Weak (RMW) 0.073 1.468
Conservative Minus Aggressive (CMA) 0.005 0.287
LIU ET AL. 7

TABLE 2 Test of abnormal return around the announcement

Three months estimation window Four months estimation window

Event Day Abnormal Return Test Statistic P‐value Abnormal Return Test Statistic P‐value
−5 −1.017 −1.303 0.170 −0.944 −1.233 0.186
−4 0.584 0.748 0.299 0.704 0.920 0.260
−3 0.557 0.714 0.239 0.830 1.084 0.141
−2 −0.620 −0.794 0.785 −0.484 −0.632 0.325
−1 −0.529 −0.678 0.315 −0.544 −0.711 0.308
0 −1.341 −1.718* 0.092 −1.322 −1.726* 0.090
1 −0.524 −0.671 0.316 −0.498 −0.651 0.321
2 −0.137 −0.176 0.391 0.070 0.092 0.396
3 0.958 1.227 0.112 1.119 1.461* 0.074
4 1.299 1.664** 0.0499 1.273 1.662* 0.0501
5 −0.683 −0.875 0.270 −0.571 −0.746 0.300

Note:
***p < 0.01, **p < 0.05, *p < 0.1.

TABLE 3 Test of abnormal return around the implementation

Three months estimation window Four months estimation window
Event Day Abnormal Return Test Statistic P‐value Abnormal Return Test Statistic P‐value

−5 −0.183 −0.382 0.369 −0.091 −0.175 0.392

−4 0.412 0.862 0.196 0.387 0.743 0.230
−3 −0.358 −0.749 0.299 −0.370 −0.710 0.308
−2 −0.273 −0.573 0.337 −0.213 −0.409 0.366
−1 −0.158 −0.332 0.376 −0.248 −0.476 0.355
0 1.035 2.167** 0.017 1.010 1.939** 0.028
1 1.257 2.633*** 0.0053 1.314 2.525*** 0.007
2 0.277 0.580 0.282 0.207 0.397 0.346
3 0.196 0.410 0.342 0.031 0.059 0.477
4 1.208 2.530*** 0.007 1.214** 2.332*** 0.011
5 0.105 0.220 0.413 −0.066 −0.128 0.394

Note:
***p < 0.01, **p < 0.05, *p < 0.1.

TABLE 4 Test of cumulative abnormal return around the announcement

Three months estimation window Four months estimation window

Event Window CAR (1) Test Statistic (2) P‐value (3) CAR (4) Test Statistic (5) P‐value (6)
(−5, 5) −1.453 −0.561 0.339 −0.367 −0.145 0.393
(0, 4) 0.255 0.146 0.393 0.642 0.375 0.370
(0, 5) −0.428 −0.224 0.387 0.071 0.038 0.397

Note:
***p < 0.01, **p < 0.05, *p < 0.1.

mentioned in more than 139 000 posts on Twitter and Facebook, positive response from the public continues to grow, it could have
and the corporation's Twitter handle @CVSExtra experienced more generated enough positive impacts on CVS’ stock to outweigh the
1
than 17 000 interactions on the day of the announcement. As the negative effects of a potential revenue loss. Thus, positive and
significant abnormal returns emerged three and four days after the
announcement. The only exception is that the abnormal returns are
1
Cigarettes Out. Health In. An analysis of the rebrand of CVS Health by Arthur found to be negative and significant on day 0, with p‐values around
W. Page Society. Available at https://docs.google.com/gview?url=http://www.
9%. Note that although abnormal returns are negative on some of the
awpagesociety.com/attachments/11d449b8506f1e3e3b52c54181a5e79a3e8be9a9/
store/29ff04562f9f89770e0fafe094f722788d62e7a3edc6a2b683aa15be84e7/ other days, they are not significant, indicating the abnormal returns
CVS‐Health‐Case‐Study.pdf Last accessed 20 April 2017. are not statistically different from zero.
8 LIU ET AL.

This negative and significant abnormal return on day 0 suggests signal sent by the CSR action is much stronger than the negative
that the initial announcement of the plan to eliminate tobacco prod- impact of tobacco revenue loss.
ucts by CVS imposed certain negative effects on CVS’ stock perfor- It is worth noting that the effects captured by the event study
mance on the same day. The shareholder theory, despite its may have been affected by two factors. First, there are two conflicting
unfavourable attention in recent literature, may play a role here. The forces that can impose opposite effects on CVS stock. The CSR action
shareholder theory suggests that firms are primarily motivated to max- itself can improve the reputation of CVS and result in a positive
imize shareholder value/wealth (Friedman, 1970). Studies adopting impact. However, the revenue loss means a negative impact which
this perspective argue that CSR spending is a cost that a firm does can partially offset the positive effects of the CSR action. Therefore,
need to bear, as it disadvantages the firm relative to its competitors the results of the events study can be regarded as the net effect of
(Jensen, 2002; McWilliams & Siegel, 1997). In the case of CVS, it could these two forces. The second factor lies in the stock market efficiency
be that the market is more concerned about the potential $2 billion in the semi‐strong form. Since CVS announced in February that it
loss of tobacco revenue, and the negative impact stemming from that intended to discontinue tobacco products starting from October, the
which outweighed the positive effects of the CSR action itself. Also, market would have priced in this information between February and
the announcement of the intention to take a CSR action doesn't nec- October. As such, one may expect less pronounced impacts on CVS’
essarily guarantee that the action will be carried out as planned. Unan- stock price in October when the plan was implemented.
ticipated events or changes in the plan can arise in between the Our analysis of cumulative abnormal returns is shown in Tables 4
announcement and the actual implementation. and 5. We consider three event windows, (−5, 5), (0, 4) and (0, 5). In
Negative effects of CVS are also documented in the literature. A the tables, we show the cumulative abnormal returns (CAR), the t‐test
study of socially responsible investing (SRI) by Renneboog, Horst, statistics, and the associated p‐values for both estimation windows,
and Zhang (2008) find that SRI funds underperform their domestic respectively. We start with the analysis of cumulative abnormal
benchmarks by −2.2% to −6.5%, based on a sample of funds from returns during the announcement period in Table 4. Although both
the USA, the UK, and many continental European and Asia‐Pacific positive and negative effects are found in the analysis of abnormal
countries. This is consistent with the conclusion that investors pay a returns on individual days, none of the cumulative returns during the
price for ethics. Furthermore, as Hong and Kacperczyk (2009) point three event windows are significant during the announcement period,
out, sin stocks usually have high expected returns and low price‐to‐ as shown by the p‐values in columns 4 and 7 in Table 4. This suggests
book or price‐to‐earnings ratios. Thus for CVS, taking out the sin com- that the overall effects of the announcement could be neutral.
ponent of its products and services may also mean lower expected Next, we move to the cumulative abnormal returns during imple-
returns. mentation (Table 5). According to the p‐values in columns 4 and 7 in
We next turn to the test on abnormal returns around the imple- Table 5, all cumulative abnormal returns are positive and significant
mentation (Table 3). Similar to the case of the announcement, we over the three event windows. Furthermore, the results become
hypothesize that the effect of the implementation is positive. We find highly significant (at 1% level) when we consider event windows
overwhelming evidence in support of Hypothesis 2, that is, positive starting from day 0. Thus, we find overwhelmingly positive and signif-
impacts on stock market performance. Based on columns 2 and 5, icant effects during the implementation of the CSR plan, based on our
abnormal returns are positive on day −4 and from day 0 to day 4 using analysis of cumulative abnormal returns. This further confirms the fact
both estimation windows. The p‐values for the t‐tests using each that the market takes it more seriously when the CSR action is actually
estimation window are reported in columns 4 and 7. We find that taking place. The results of cumulative abnormal returns reinforce our
the positive abnormal returns are significant on day 0 (5% level), day findings in the daily abnormal returns analysis and provide additional
1 (1% level), and day 4 (1% with a three‐month estimation window evidence in support of Hypothesis 2, the positive effects of
and 5% with a four‐month estimation window). Such consistent implementation.
results have several implications. First, the market values the imple- To check the robustness of our results, we also use the single fac-
mentation of the CSR action more than the announcement of the plan. tor market model to calculate the abnormal returns. It has been
Business plans, although announced, can change before actually being discussed in the literature that the coefficients in the simple model
implemented. The actual implementation of the tobacco withdrawal is are generally statistically significant and thus the calculated abnormal
a further confirmation of CVS’ determination to be socially responsi- returns are highly reliable; other more sophisticated methods lack such
ble. Second, the positive abnormal returns indicate that the positive advantages and rarely improve the fit of the model (Becchetti et al.,

TABLE 5 Test of cumulative abnormal return around the implementation

Three months estimation window Four months estimation window

Event
Window CAR (1) Test Statistic (2) P‐value (3) CAR (4) Test Statistic (5) P‐value (6)
(−5, 5) 3.518 2.221** 0.036 3.174 1.838* 0.075
(0, 4) 3.973 3.721*** 0.001 3.775 3.243*** 0.003
(0, 5) 4.078 3.486*** 0.001 3.708 2.908*** 0.007

Note:
***p < 0.01, **p < 0.05, *p < 0.1.
LIU ET AL. 9

2012; Brown & Warner, 1985; Campbell, Lo, & McKinlay, 1997). The and the Rite Aid dummy. Their coefficients measure the comparison
single factor market model takes the following specification: of the long‐term monthly returns between CVS and its two compet-
itors. A positive coefficient would suggest that the long‐term
Rt ¼ α þ βRet þ εt (8)
monthly return of the specific company (Walgreen or Rite Aid) is
on average higher than that of CVS. As depicted by column 1,
We test both the single day abnormal returns and cumulative
although both coefficients are negative (−0.266 and − 0.023), they
abnormal returns. Results are shown in the Appendix Tables A1–A4.
are not significant at the conventional levels. Thus, the CVS’ monthly
Overall, we obtain very similar results to those shown in Tables 2–5.
return is not significantly different from its competitors in the long
According to Tables A1 and A2, positive abnormal returns are found
term.
on days 4 and 5 during the announcement (Table A1) and days 0, 1
Jensen's alpha approach also confirms the same finding, as shown
and 4 during the implementation (Table A2). In addition, the cumula-
in column 3. The key variable of interest here is Jensen's alpha, repre-
tive abnormal returns are overwhelmingly positive during the imple-
sented by the y‐intercept. The coefficient of the variable is 0.039.
mentation as shown in Table A4, but not during the announcement
However, the coefficient is not statistically significant.
as presented in Table A3. The only major difference between the
Next, we examine results from the fixed‐effects panel data model,
results using Fama–French five‐factor model (Table 1) and the results
which are reported in column 4 of Table 6. The variable of interest is
using the single factor model (Table A1) is that the negative abnormal
the dummy variable that presents the CVS return during the post‐
return on day 0 of the announcement becomes insignificant using the
event period, CVS post‐event dummy. The coefficient is negative
single factor market model.
(−0.009) but insignificant, suggesting that the post‐event monthly
Overall, our analysis shows consistent evidence in support of
returns of CVS stock are not substantially higher or lower. This further
Hypothesis 1 and Hypothesis 2. We find the abnormal returns are
supports our findings in the first two models.
positive and statistically significant on day 3 and day 4 during the
Based on the three methods we employed in the long‐term
announcement, and day 1, day 3 and day 4 during the implementation.
analysis, we do not find evidence in support of Hypothesis 3.
Our analysis of cumulative abnormal returns provides further
Instead, we find that the long‐term effect is neutral. We believe
evidence.
that the lack of evidence for the long‐term hypotheses does not
necessarily mean our study contradicts the theory or previous find-
4.2 | Results of long‐term analysis ings of positive effects. It is instead a result of the interaction
The results of the long‐term analysis are summarized in Table 6. We between the opposing forces due to the uniqueness of our case.
consider three alternative models for our long‐term analysis. Column While being socially responsible itself means a positive impact for
2 shows the results of the pooled OLS estimate, which is equivalent CVS, the tobacco revenue loss, on the other hand, can impose neg-
to a modified BHAR method. Column 3 presents the results of ative effects. The initial positive impact could be the net effect of
Jensen's alpha approach while column 4 shows results using a fixed‐ the trade‐off between the two forces. In comparison to the initial
effects panel data model. event period, the potential loss of tobacco revenue is realized and
We first examine results obtained using the pooled OLS method, factored into the stock performance in the long term. Actually, after
as shown in column 1. The variables of interests are the two dummy the ban on tobacco in October 2014, front‐of‐store sales were
variables that represent the two competitors, the Walgreen dummy reported to have decreased at various times in 2015. For example,

TABLE 6 Long‐term analysis

Variables Pooled OLS (1) Jensen's Alpha Model (2) Fixed‐Effects Panel Data Model (3)

Walgreen Dummy −0.266 (0.149)

Rite Aid Dummy −0.023 (0.027)
Jensen's alpha 0.039 (0.049)
CVS post‐event Dummy −0.009 (0.011)
Market Return 0.013 (0.005) 0.004** (0.002) 0.012*** (0.003)
SMB 0.003 (0.005) −0.001 (0.004) 0.003 (0.004)
HML 0.006 (0.003) −0.008** (0.003) 0.004* (0.002)
Momentum 0.008 (0.003) 0.007*** (0.002)
CMA 0.011 (0.007) 0.0101** (0.004)
Book‐to‐Market Ratio −0.009 (0.004) −0.001*** (0.0004)
UMD −0.003 (0.003)
Number of Observations 84 25 147
R‐squared 0.246 0.3 0.22

Note: 1. Robust standard errors are included

2.
***p < 0.01, **p < 0.05, *p < 0.1.
10 LIU ET AL.

according to Bloomberg News, “front‐of‐store sales for locations its revenue, the positive signal sent by the action was well received
that have been open at least a year dropped 7.2%, reflecting the by the market, investors and the public. Thus, the overall reaction from
absence of tobacco products” (Cortez, 2015). In August of 2015, the stock market is positive during the implementation of the CSR
CNN reported a decline of almost 8% in general merchandise sales action. Future policies can be directed to encourage companies to take
by CVS (Egan, 2015). Thus, the positive effects of CSR could have CSR actions that may be costly but which can bring valuable benefits
been offset by the negative effects of the lost revenue in the long to the public, including economic, health and environmental benefits.
term. Meanwhile, companies can still gain in the stock market, as our study
shows.
Third, the positive stock market reactions revealed provide an
5 | D I S C U S S I O N A N D CO N C L U SI O N important example for companies that want to consider CSR more
broadly, including CSR actions that may impose constraints on its
This paper contributes to the literature on the effect of CSR on revenues. Analysts had projected a domino effect as several retail
firm stock market performance by examining a special case of pharmacy players were expected to follow CVS’ lead. However,
CSR that involves both a firm's socially responsible action and its according to Ward, Roy, and Edmondson (2016), the reactions to
significant direct impact on firm revenue. We investigate how CVS’ move were mixed. Walgreens refused to follow CVS’ path and
CVS’ elimination of tobacco products impacts its stock market continued the sale of tobacco in its stores (Japsen, 2014). Costco indi-
performance in the short and the long term. Our results show that cated that it would slowly scale down the number of tobacco products
the stock market reacted positively to both the announcement and in its stores (Ward et al., 2016). It is possible that these companies are
the implementation of the CSR action in the short term. We also more concerned about the potential revenue loss and intend to evalu-
find that abnormal returns during the implementation event days ate the impact of CVS’ action before making any commitment. The
are more significant than those during the announcement event findings of this study can provide information about what they can
days. Furthermore, our study of cumulative abnormal returns shows expect from the public and investors. Also, although the two biggest
strong evidence of positive effects of the implementation. When competitors of CVS, Walgreen and Rite Aid, didn't follow CVS’ lead
we extend our analysis to the long‐term period, the positive in withdrawing tobacco, the results of this study can provide them
effects captured in the short term vanish and the net effects are with important insights if they do want to consider undertaking it or
neutral. similar CSR actions in the future.
This research extends to prior studies on the relationship between More broadly, our results have important implications for com-
CSR actions and firm performance as we examine the wider connec- panies in every industry, not just the retail pharmacy industry:
tions between corporate social activities with substantial costs and socially responsible actions, such as the elimination of harmful
overall stock market performance. The findings of positive impact in products (e.g. tobacco products), socially responsible marketing,
the short term provide further evidence in support of an overall responsible production and sustainable consumption, could impose
favourable attitude towards CSR activities. substantial revenue losses or extra expenditure, but may also have
This study provides important implications for CSR theorists, positive and significant impacts on corporate financial performance.
practitioners, investors and policymakers. First, from a theoretical per- For instance, corporate actions that enhance product quality, ulti-
spective, the findings of positive financial performance support and mately attract and retain customer loyalty; corporate measures that
extend the shareholder perspective that actions taken to advance ensure responsible marketing policies are established and adhered
the interest of legitimate corporate stakeholders have a favourable to; and the pursuit of responsible production and consumption poli-
outcome for the organization (Margolis & Walsh, 2003; Orlitzky cies that ensure environmental sustainability for current and future
et al., 2003). Our findings, therefore, add to and strengthen existing generations must be enacted and implemented consistently. These
research results regarding the relationship between CSR and financial actions benefit not only the stakeholders, but also shareholders,
performance. Our results also suggest that CSR actions with predict- and are consistent with both the stakeholder theory and shareholder
able costs do not necessarily conflict with the interests of share- theory.
holders. Actually, the positive effects suggest such actions can In particular, the overwhelmingly positive impacts found during
benefit both stakeholders and shareholders simultaneously. Such the implementation provides companies with more incentives to con-
types of corporate strategic behaviour can bring about a win‐win sider undertaking similar CSR actions. This, in our view, is a significant
situation. contribution of this study. Furthermore, this study extends prior
Second, the CSR action taken by CVS turns out to be beneficial to knowledge of what would typically be considered as CSR, and gives
both the company and the public. The withdrawal of tobacco, and the managers an additional dimension to consider when they attempt to
provision of support services to help people stop smoking, can be of conceptualize and operationalize CSR strategies that fit their business
tremendous benefit to public health. It can discourage tobacco use, circumstances.
promote a tobacco‐free environment and reduce tobacco‐induced dis- Fourth, given the fierce competition in the pharmaceutical retail
eases, which can be fatal and costly. According to CDC, the annual market, it is critical for a firm to build company reputation to differen-
total economic cost of smoking in the USA is over $300 billion. Such tiate itself from the competitors. The action taken by CVS represents
an action by CVS means huge savings to society and improved health such an endeavour to improve the company's brand. The heated dis-
outcomes. On the other hand, although CVS faces a $2 billion loss in cussion about CVS on social media (such as Facebook and Twitter)
LIU ET AL. 11

was a very effective way for CVS to enhance its reputation among the Brammer, S., & Millington, A. (2008). Does it Pay to be Different? An
public. This study provides evidence of the effectiveness of such Analysis of the Relationship between Corporate Social and Financial
Performance. Strategic Management Journal, 29(12), 1325–1343.
endeavours and thus enables decision‐makers to make better
Brown, S. J., & Warner, J. B. (1985). Using daily stock returns: the case of
informed decisions.
event studies. Journal of Financial Economics, 14(1), 3–31.
This study also has certain limitations. First, we only analysed
Campbell, J. Y., Lo, A. W., & McKinlay, C. (1997). The econometrics of
CVS’ stock performance when examining the effects of CVS’ CSR financial markets. Princeton, NJ: Princeton University Press.
action. Walgreen and Rite Aid are close competitors of CVS in that Carroll, A. (1991). The pyramid of corporate social responsibility: Toward
those companies share very similar store structures and locations, the moral management of organizational stakeholders. Business Hori-
zons, 34, 39–48.
etc. It would be useful to examine Walgreen's and Rite Aid's stock
market performances during the same period to form a complete pic- Centers for Disease Control and Prevention (2017). Data and Statistics:
Fast Facts and Fact Sheets. Available at https://www.cdc.gov/
ture of the total effect of CVS’ tobacco withdrawal. Previous empirical tobacco/data_statistics/fact_sheets/index.htm Accessed 15 Septem-
CSR studies use different indicators of financial performance, includ- ber 2017.
ing accounting‐based measures such as profits, returns on assets, Cho, S., Lee, C., & Pfeiffer, R. (2013). Corporate Social Responsibility Per-
and market‐based measures such as stock returns. In our study we formance and Information Asymmetry. Journal of Accounting and
Public Policy, 32(1), 71–83.
selected market‐based measures. To make the results comparable to
Clarkson, P. M., Fang, X., & Richardson, G. (2013). The Relevance of Envi-
previous research, it may also be helpful to examine the financial per-
ronmental Disclosures: Are such disclosures incrementally
formance of CVS by adopting accounting‐based measures. Future informative? Journal of Accounting and Public Policy, 32(5), 410–431.
research could provide these insights. Cochran, P. L., & Wood, R. A. (1984). Corporate social responsibility and
Second, in the long‐term part of our study, we were only able to financial performance. Academy of Management Journal, 27(1), 42–56.
examine the effects within two years of the implementation of the Cortez, M. (2015). Quitting Smoking Has Been Just Fine for CVS.
tobacco ban. Thus, the long‐term impact is still inconclusive. Studies Bloomberg. Available at https://www.bloomberg.com/news/articles/
2015‐02‐10/cvs‐profit‐beats‐estimates‐in‐first‐full‐tobacco‐free‐quar-
of the long‐term effects after 3–5 years of the CSR action may pro-
ter Last accessed 24 March 2017.
vide additional information. Our study shows no long‐term effects
De Bakker, F. G., Groenewegen, P., & Den Hond, F. (2005). A bibliometric
within two years of CVS’ tobacco withdrawal. However, as CVS con- analysis of 30 years of research and theory on corporate social respon-
tinues to build its brand and company image, it may gain more from sibility and corporate social performance. Business & Society, 44(3),
283–317.
cultivating its image of being socially responsible. With the negative
Donaldson, T., & Preston, L. E. (1995). The Stakeholder Theory of the
effects of tobacco revenue loss being diluted, the overall long‐term
Corporation: Concepts, Evidence, and Implications. Academy of
effects may change. Future studies are needed to verify the long‐term Management Review, 20, 65–91.
effects over a longer period. Finally, our study focuses on a single Ducassy, I. (2013). Does Corporate Social Responsibility Pay Off in Times
company and a single event. More studies of similar CSR actions that of Crisis? An Alternate Perspective on the Relationship between Finan-
bear huge costs are needed to further confirm the overall effects of cial and Corporate Social Performance. Corporate Social Responsibility
and Environmental Management, 20, 157–167.
this type of CSR action.
Egan, M. (2015). CVS banned tobacco. Now its sales are hurting. CNN.
Available at: http://money. cnn.com/2015/08/04/investing/cvs‐earn-
ORCID ings‐cigarettes/Last accessed 3/24/2017.
Fifka, M. S., & Berg, N. (2014). Managing Corporate Social Responsibility
Michael Opara http://orcid.org/0000-0001-7441-7026
for The Sake of Business and Society. Corporate Social Responsibility
and Environmental Management, 21, 253–257.
RE FE R ENC E S
Flammer, C. (2015). Does corporate social responsibility lead to superior
Aupperle, K. E., Carroll, A. B., & Hatfield, J. D. (1985). An empirical financial performance? A regression discontinuity approach. Manage-
examination of the relationship between corporate social responsi- ment Science, 61(11), 2549–2568.
bility and profitability. Academy of Management Journal, 28(2),
Fombrun, C. J. (1996). Reputation: Realizing Value from the Corporate Image.
446–463.
Boston, MA: Harvard Business School Press.
Banerjee, S., & Wathieu, L. (2017). Corporate Social Responsibility and Freeman, R. E., Harrison, J. S., & Wicks, A. C. (2007). Managing for stake-
Product Quality: Complements or Substitutes? International Journal of holders: survival, reputation, and success. New Haven, CT: Yale
Research in Marketing, 34(3), 734–745. University Press.
Barnett, M. L., & King, A. A. (2008). Good Fences Make Good Neighbors: A Freeman, R. E., Harrison, J. S., Wicks, A. C., Parmar, B. L., & de Colle, S.
Longitudinal Analysis of an Industry Self‐Regulatory Institution. Acad- (2010). Stakeholder theory: the state of the art. New York, USA:
emy of Management Journal, 51, 1150–1170. Cambridge University Press.
Barnett, M. L., & Salomon, R. M. (2012). Does it Pay to be Really Friedman, M. (1970). The social responsibility of business is to increase its
Good? Addressing the Shape of the Relationship between Social profits. The New York Times [13 September 1970].
and Financial Performance. Strategic Management Journal, 33, Frynas, J. G., & Yamahaki, C. (2016). Corporate Social Responsibility:
1304–1320. Review and Roadmap of Theoretical Perspective. Business Ethics: A
Beare, D., Buslovich, R., & Searcy, C. (2014). Linkages between Corporate European Review, 25, 258–285.
Sustainability Reporting and Public Policy. Corporate Social Responsibil- Garriga, E., & Mele, D. (2004). Corporate Social Responsibility Theories:
ity and Environmental Management, 21, 336–350. Mapping the Territory. Journal of Business Ethics, 53, 51–74.
Becchetti, L., Ciciretti, R., Hasan, I., & Kobeissi, N. (2012). Corporate social Ghoul, S., Guedhami, O., Kwok, C. C. Y., & Mishra, D. R. (2011). Does cor-
responsibility and shareholder's value. Journal of Business Research, 65, porate social responsibility affect the cost of capital? Journal of Banking
1628–1635. & Finance, 35, 2388–2406.
12 LIU ET AL.

Gibson, K. (2000). The moral basis of stakeholder theory. Journal of Busi- Michelon, G., Boesso, G., & Kumar, K. (2013). Examining the Link
ness Ethics, 26(3), 245–257. between Strategic Corporate Social Responsibility and Company
Performance: An Analysis of the Best Corporate Citizens. Corporate
Gomez‐Bezares, F., Przychodzen, W., & Przychodzen, J. (2016). Bridging
Social Responsibility and Environmental Management, 20, 81–94.
the Gap: How Sustainable Development can help Companies create
Shareholder Value and Improve Financial Performance. Business Ethics: Miras‐Rodriguez, M., Carrasco‐Gallego, A., & Escobar‐Perez, B. (2015). Are
A European Review, 26, 1–17. Socially Responsible Behaviours Paid Off Equally? A Cross‐cultural
Analysis. Corporate Social Responsibility and Environmental Management,
Hoffman, A. J., & Ventresca, M. J. (2002). Organizations, policy, and the nat-
22, 237–256.
ural environment: institutional and strategic perspectives. Redwood City,
CL: Stanford University Press. Mitchell, M., & Stafford, E. (2000). Managerial decisions and long‐term
stock price performance. Journal of Business, 73, 287–329.
Hong, H., & Kacperczyk, M. (2009). The price of sin: The effects of social
norms on markets. Journal of Financial Economics, 93, 15–36. Oberseder, M., Schlegelmilch, B. B., & Murphy, P. E. (2013). CSR Practices
and Consumer Perception. Journal of Business Research, 66, 1839–1851.
Howe, J., & Schlarbaum, G. (1986). SEC Trading Suspensions: Empirical
OECD (2008). Promoting Sustainable Consumption: Good Practices in
Evidence. The Journal of Financial and Quantitative Analysis, 21,
OECD Countries. Available at: https://www.oecd.org/greengrowth/
323–333.
40317373.pdf Last accessed 12 April 2018.
Huang, X., & Watson, L. (2015). Corporate Social Responsibility Research
Orlitzky, M., Schmidt, F. L., & Rynes, S. L. (2003). Corporate social and
in Accounting. Journal of Accounting Literature, 34, 1–16.
financial performance: a meta‐analysis. Organization Studies, 24(3),
Jahdi, K. S., & Acikdilli, G. (2009). Marketing Communications and Corpo- 403–441.
rate Social Responsibility (CSR): Marriage of Convenience or Shotgun Orlitzky, M., Siegel, D. S., & Waldman, D. A. (2011). Strategic Corporate
Wedding? Journal of Business Ethics, 88, 103–113. Social Responsibility and Environmental Sustainability. Business & Soci-
Japsen, B. (2014) Why Walgreen won't stop selling tobacco like CVS ety, 50(1), 6–27.
Health. Forbes. Retrieved from: https://www.forbes.com/sites/ Peters, R., & Mullen, M. (2009). Some Evidence of the Cumulative Effects
brucejapsen/2014/09/04/why‐walgreen‐wont‐stop‐selling‐tobacco‐ of Corporate Social Responsibility on Financial Performance. Journal of
like‐cvs‐health/#68fcbc7141a4 Last accessed 24 March, 2017. Global Business Issues, 3(1), 1–14.
Jensen, M. C. (2002). Value maximization, stakeholder theory, and the cor- Porter, M. E., & Kramer, M. R. (2006). Strategy & society: the link between
porate objection function. Business Ethics Quarterly, 12, 235–256. competitive advantage and corporate social responsibility. Harvard
Jones, T. (1995). Instrumental stakeholder theory: A synthesis of ethics and Business Review, 2006, 78–92.
economics. Academy of Management Review, 20(2), 404–437. Renneboog, L., Horst, J. T., & Zhang, C. (2008). The price of ethics and
stakeholder governance: the performance of socially responsible
Kim, J. (2010). Assessing the Long‐term Financial Performance of Ethical
mutual funds. Journal of Corporate Finance, 14, 302–322.
Companies. Journal of Targeting, Measurement and Analysis for Market-
ing, 18(3–4), 199–208. Serra, A. P. (2002). Event study tests: A brief survey. Working paper. Avail-
able at: http://www.fep.up.pt/investigacao/workingpapers/wp117.pdf
Kothari, S. P., & Warner, J. B. (2007). Econometrics of event studies. In
Last accessed 8 March 2017.
Handbook of corporate finance (Vol. 1) (pp. 3–36). Amsterdam, The
Netherlands: Elsevier. Servaes, H., & Tamayo, A. (2013). The impact of corporate social responsi-
bility on firm value: The role of customer awareness. Management
Kryzanowski, L., & Nemiroff, H. (2001). Market Quote Behaviour and Science, 59(5), 1045–1061.
Spread Component Cost Behaviour Around Trading Halts for Stocks
Realised on the Montreal and Toronto Stock Exchanges. The Financial Tenbtermsel, A. E. (2007). Today's ethical issues: A perspective from
Review, 37, 115–138. accounting, finance, management, and marketing. Journal of Business
Ethics, 70(1), 1–3.
Landrum, N. E., & Ohsowski, B. (2018). Identifying Worldviews on Corpo-
The Economist (2008). A survey of corporate social responsibility: just good
rate Sustainability: A Content Analysis of Corporate Sustainability
business http://www.economist.com/node/10491077[17 January 2008].
Reports. Corporate Social Responsibility and Environmental Management,
27, 128–151. Torelli, C. J., Basu Monga, A., & Kaikati, A. M. (2012). Doing poorly by
doing good: Corporate social responsibility and brand concepts. Journal
Lys, T., Naughton, J. P., & Wang, C. (2015). Signaling through corporate
of Consumer Research, 38(5), 948–963.
accountability reporting. Journal of Accounting and Economics, 60(1),
56–72. Vogel, D. J. (2005). Is there a Market for Virtue? The Business Case for
Corporate Social Responsibility. California Management Review, 47(4),
Margolis, J., Elfenbein, H., & Walsh, J. (2009). Does it pay to be good? A 19–45.
meta‐analysis and redirection of research on the relation between cor-
Waddock, S. A., & Graves, S. B. (1997). The corporate social performance‐
porate social and financial performance. Working Paper, Harvard
financial performance link. Strategic Management Journal, 18, 303–319.
University.
Ward, C. B., Roy, D. P., & Edmondson, D. R. (2016). Is CVS just ‘blowing
Margolis, J. D., & Walsh, J. P. (2003). Misery loves companies: Rethinking
smoke?’ Evaluating the CVS decision to ban tobacco products. Case
social initiatives by business. Administrative Science Quarterly, 48(2),
Studies in Strategic. Communication, 5, 249–264.
268–305.
Zucker, L. G. (1983). Organizations as Institutions. In S. B. Bacharach (Ed.),
Marti, P. L., Rovira‐Val, M. R., & Drescher, L. G. J. (2015). Are Firms that
Perspectives in Organizational Sociology: Theory and Research. Green-
Contribute to Sustainable Development Better Financially? Corporate
wich, CT: JAI Press.
Social Responsibility and Environmental Management, 22, 305–319.
Zucker, L. G. (1987). Institutional Theories of Organizations. Annual Review
McWilliams, A., & Siegel, D. (1997). The role of money managers in of Sociology, 13, 443–464.
assessing corporate social responsibility research. Journal of Investing,
6, 98–107.
McWilliams, A., Siegel, D., & Wright, P. M. (2006). Corporate Social How to cite this article: Liu L, Li S, Opara M. Corporate social
Responsibility: Strategic Implications. Journal of Management Studies,
responsibility and strategic company behaviour: CVS Health's
43(1), 1–18.
discontinuation of tobacco products. Corp Soc Resp Env Ma.
Merlo, L. J. (2014). Message from Larry Merlo, President and CEO. CVS
Health. Retrieved from https://cvshealth.com/thought‐leadership/ 2018;1–13. https://doi.org/10.1002/csr.1639
expertvoices/message‐from‐larry‐merlo‐president‐and‐ceo
LIU ET AL. 13

APPENDIX A
TABLE A.1 Test of abnormal return around the announcement: single factor model

Three months estimation window Four months estimation window

Event Day Abnormal Return Test Statistic P‐value Abnormal Return Test Statistic P‐value
−5 −1.146 −1.283 0.174 −0.949 −1.071 0.223
−4 −0.667 −0.747 0.300 −0.629 −0.710 0.308
−3 0.365 0.409 0.342 0.534 0.603 0.274
−2 −1.258 −1.408 0.919 −0.963 −1.087 0.220
−1 −0.297 −0.332 0.376 −0.228 −0.257 0.385
0 −1.066 −1.193 0.194 −0.927 −1.047 0.229
1 −0.081 −0.091 0.396 −0.043 −0.048 0.397
2 −0.693 −0.777 0.293 −0.663 −0.749 0.300
3 0.415 0.465 0.322 0.527 0.594** 0.277
4 1.717 1.922** 0.0290 1.763 1.991** 0.0249
5 −0.635 −0.711 0.308 −0.520 −0.588 0.334

Note:
***p < 0.01, **p < 0.05, *p < 0.1.

TABLE A.2 Test of abnormal return around the implementation: single factor model

Three months estimation window Four months estimation window

Event
Day Abnormal Return Test Statistic P‐value Abnormal Return Test Statistic P‐value
−5 0.284 0.545 0.342 0.249 0.440 0.361
−4 0.441 0.847 0.200 0.270 0.476 0.318
−3 −0.230 −0.440 0.360 −0.262 −0.463 0.357
−2 −0.057 −0.109 0.395 −0.150 −0.264 0.384
−1 −0.104 −0.200 0.389 −0.208 −0.366 0.372
0 0.861 1.651* 0.052 0.702 1.239 0.109
1 1.063 2.039** 0.023 0.991 1.749** 0.042
2 0.565 1.084 0.141 0.546 0.963 0.169
3 0.319 0.612 0.271 0.224 0.395 0.347
4 1.214 2.328** 0.012 1.045 1.845** 0.034
5 0.204 0.390 0.349 0.219 0.386 0.369

Note:
***p < 0.01, **p < 0.05, *p < 0.1.

TABLE A.3 Test of cumulative abnormal return around the announcement: single factor model

Three months estimation window Four months estimation window

Event
Window CAR Test Statistic P‐value CAR Test Statistic P‐value
(−5, 5) −3.345 −1.129 0.209 −2.098 −0.714 0.307
(0, 4) 0.291 0.146 0.393 0.657 0.332 0.376
(0, 5) −0.344 −0.157 0.392 0.657 0.332 0.376

TABLE A.4 Test of cumulative abnormal return around the implementation: single factor model

Three months estimation window Four months estimation window

Event
Window CAR Test Statistic P‐value CAR Test Statistic P‐value
(−5, 5) 4.560 2.637** 0.014 3.625 1.930* 0.063
(0, 4) 4.021 3.450*** 0.002 3.507 2.769** 0.010
(0, 5) 4.225 3.309*** 0.002 3.726 2.685** 0.012

Note:
***p < 0.01, **p < 0.05, *p < 0.1.